Faculté des sciences économiques et sociales, Wirtschafts- und Sozialwissenschaftliche Fakultät

Starting with the Internet of Things, attempts have already been made to propose common vocabularies and models in order to address and combine things (e.g. tagged objects, smart-phones, ...), their human owners and some services and associated resources within a generic architecture (IoT-A European project). Our goal is (1) to extend this effort to propose a model including complex (even delayed and decomposable) services, as well as some useful semantic information (context) and "alerts" (callbacks) from the servers to the clients; (2) to validate our model with prototypes implemented by our group or in collaboration with other institutions (e.g. SAP Research Switzerland, the University of Applied Science of Western Switzerland, the WNEC Lab in Taiwan) and which will be based on RESTFul principles; (3) and finally to abstract from our experiences tools, design patterns and eventually a whole framework for facilitating the creation of WoR worlds.

The presence of software systems in every aspect of our life results in increased requirements and expectations addressed by new development projects. There is for instance a progressing need to optimise interaction of users with services and devices, so to achieve efficiently a specific goal. The domain of home computing constitutes a typical example of an environment which is characterized by complex requirements with regard to context awareness, intelligent assistance, autonomy, dynamic discovery, dynamic service composition, and easy services management for end users. Innovative development paradigms have been suggested to meet these new demanding conditions. One of those paradigms is Services Oriented Computing [Wik10] which has been proposed as a new solution for heterogeneous and changing environments. Of course, the industry requires new developed tools to support these approaches. Web Services [Wik09] technologies are becoming a de facto standard to integrate distributed applications and systems. Web services are the preferred standards-based way to realize service oriented architecture computing. However, developing applications that support web services interfaces will not be enough to provide complete and coordinated business processes. Web Services Business Process Execution Language (BPEL4WS [IBM02]) has been proposed as a standard solution for centralised Web service orchestration. Unfortunately, BPEL4WS is rather static and it does not deal with changing environments where services and devices might change their addresses or be replaced. Additionally, web services have to be known already during the process design phase. This prevents the user from selecting another service afterwards, when many different services are available to provide similar functionality components. Thus, we need new approaches to help avoid these limitations. This work aims to propose an approach based on semantic description and ontologies in order to discover and compose services in a changing environment. It is not necessary to know the exact service at process design time. Furthermore, our goal is to offer a partial automation of service composition, with a human controller. We propose to enable customers to select and configure available services to meet their requirements and reach a specific goal. We have the objective to explore how ontology can help making intelligent-assistant and goal oriented composition achievable.

In the recent years, the so-called Internet of Things (IoT) attracted much attention and now begins to gain widespread acceptance. Beside the IoT another research focus of the future internet is the internet of Services (IoS). We believe that a combination of these two powerful research areas will have far-reaching consequences, not only from a social, but also from a business point of view. This research project contributes to overcome existing obstacles concerning the application of IoT-technologies within business processes.

Social networks and communities have become an important environment for exchanging information about products, services, music, and movies, among other things. In an information and knowledge society, such technologies could also improve democratic processes, increase citizens? interest in political issues, enhance participation, and renew civic engagement. However, the difficulty of finding other citizens or parties that share common goals is still a barrier. In this PhD thesis, a fuzzy-based recommender system architecture for stimulating political participation and collaboration is proposed.

The fuzzy logic theory is based on intuitive reasoning and takes into account human subjectivity and imprecision. Unlike classical data mining techniques such as cluster or regression analysis, fuzzy logic enables the use of non-numerical values and introduces the notion of linguistic variables. Using linguistic terms and variables hides the complexity of the application domain and permits a more intuitive and human-oriented querying process. A fuzzy classification approach with its query facility improves customer equity, enables loyalty programs, automates mass customization issues, and refines marketing campaigns.

Recent trends in ubiquitous and pervasive computing, together with new sensor technologies, wireless communication standards, powerful mobile devices and wearable computers strongly support novel information management applications in healthcare. In addition, these technologies also allow making this information ubiquitously available on different types of devices to different types of users (physicians, nurses, or patients), independent of their current location. Both trends together will significantly improve the quality of treatment and care for patients and the elderly. If we consider our aging society, the amount of elderly people suffering from one or more chronic diseases is increasing. Telemonitoring applications in home care enable healthcare institutions to take care of their patients while they are out of hospital. This is especially useful for managing various chronic diseases as well as for measuring the effects of treatments under real-life conditions. Given appropriate sensor technology and a reliable infrastructure for telemonitoring and data stream processing, i.e., an infrastructure their users can count on, caregivers will decide to equip their patients with a wearable telemonitoring system consisting of, e.g., ECG, blood pressure, and oxygen saturation sensors attached to patient?s body. Context information about the patient like his current activity or position, and body sensor information is processed by a telemonitoring infrastructure. It will analyze the data accumulated, extract and forward relevant information to the care provider in charge. In this scenario, disease management is improved by detecting critical situations that might occur between consultations of the physician. In case of relevant changes of the patient?s health condition, his/her physician in charge or the emergency service will be automatically informed, and is able to retrieve all medical important data. Similar technologies are also needed for applications in stationary care. Consider, for instance, a doctor on a ward round who is equipped with a PDA for accessing the electronic health record of a patient she is about to see. Due to the display limitations of these devices, it would be desirable to dynamically switch to a more powerful display (e.g., monitor in the examination room) whenever this is needed, for instance for the display of high-resolution X-Ray images. Another important aspect is the adaptation to different end users. The electronic health record could provide a special interface for doctors and another one for nurses. Using this interface adaptation makes it easier and faster for a nurse and a doctor to fill in and access the needed information. Such applications require an infrastructure that is able to dynamically adapt to the particular context and location of a user. This includes the handling of interfaces (sessions that need to be transferred from one device to another) but also the location and context-dependent combination, processing, and managing of data, in particular of continuous data streams coming from the different soft- and hardware sensors over heterogeneous devices like body sensors integrated in embedded systems, mobile devices, like PDAs or smart phones, which allow for patient mobility, and stationary PCs for further processing and long term storage connected. This location and context-aware adaptation must be done in a user-friendly way and has to be done on top of a highly dependent and reliable infrastructure which can potentially be life-saving.

Due to information overload, it is important to know how to extract relevant information. This leads to the question: how can relevant information be found? The problem with representing documents and queries through sets of keywords is that they yield results which are only partially relevant to their actual semantic contents. As a consequence, the matching of a document to the query terms is vague. This can be clarified by considering that every query expression defines a fuzzy set and that each document has a degree of relationship in this set. To search for information from weblogs, a user types search terms and related relevance (user need) in an interface. This user need will be sent to the query engine. The query engine generates a search query and sends it to a meta search engine. The meta search engine then sends user requests to numerous other search engines and aggregates the results into a single list. The idea behind this is to bring these components together and create a search engine for weblogs with the use of fuzzy methods to search for related documents.

This research project aims at the automated definition of fuzzy classes for marketing management using inductive methods. First, new algorithms for membership function induction are developped and evaluated. Second, a prototype implementation is programmed. And third, an application framework for marketing analytics is designed, in order to define fuzzy target groups for individual marketing campaigns.

Many questions in sophisticated management decisions cannot be answered simply with ''yes'' or ''no''. In reality, those questions will be answered like ''yes, but with reservation...'' or ''in this case no, but in this case yes...''. Most of the time it is an assessment of different impacts, which leads to imprecise or fuzzy answers. Additionally, the bases for decisions are not always quantitative (for instance monetary) but qualitative information. In order to handle qualitative and imprecise data, the theory of fuzzy sets can be applied. Fuzzy sets provide mathematical meanings to natural language and are therefore able to handle the imprecision of the natural language. As a Data Warehouse often provides the base of information for taking business decision, it is crucial that it is able to challenge imprecise and qualitative information. Today's Data Warehouse architectures only hardly handle imprecise and qualitative data. The fuzzy Data Warehouse (fDWH) extends a common Data Warehouse architecture with the ability to handle fuzzy facts (the data tuple in the Data Warehouse) and fuzzy dimensions. To build a fDWH an equivalence class model with linguistic variables an contexts for the facts has to be created. For instance: The fact turnover is defined as a set [0,...,1000], what represents the value rane. It has equivalence classes [0,...600] and [400,...,1000]. It is now possible to define two contexts ''small'' and ''big'' for the linguistic variable ''turnover'', whereas small is equivalent to [0,...,600] and big to [400,...,1000]. For each value of the fact turnover it is now possible to define a degree of membership to each context. Extending the dimensions of the fuzzy Data Warehouse with fuzzy sets allow handling imperfect data. Therefore dimensions can be defined as fuzzy decompositions. Missing values can be more easily predicted comparing the membership degrees of neighbor fact values. This leads to a more flexible Data Warehouse that handles more efficient fuzzy, imperfect and context dependent data.

DOME allows the specification, reuse and modification of facts bases, scenarios, decision support tasks, exogenous decision sets in order to speed up and enhance cooperative decision support. The underlaying decision space is represented by a knowledge base consisting of a collection of mathematical models and data bases. Until 2004 modifications of the decision space were limited to re-configuering modules of the modelbase. The introduction of new knowledge and the modification of exisiting chunks of knowledge was reserved to modelers using special modeling tools; it was hidden from DSS users. This project aims to integrate a Model Manager into DOME in order to include modeling and sensitivity analysis of model instances within the use cases of a typical DSS user (i.e. a Decision Assistant).

SME often lack of the resources to implementing an own Webshop. However, they too could participate in the growing ebusiness market. This project aims to offer an easy to implement Web Store. The graphical front-end allows a comfortable configuration. With the help of the many pre-configured skins and the ability to have several languages on the client side as well as on the administrator side the enterprise is able to set up its individual shop without any special knowledge.

The aim of this personal research is to develop decision support tools for small and medium enterprises using operational reseach methods and techniques. The research field covers different aspects of the industrial organisation : production system layout and design, planning, scheduling, real time control, logistics, timetabling, etc.

The aim of this project is to develop an abstract and general theory of information systems. This theory leads to efficient generic computing structures. In particular, such structures are useful in the case of distributed systems.

Various themes including graph coloring, integer polyhedra, total dual integrality, total dual conformity, polarity, submodular functions.